How to check your Codex usage limits

The quick answer: /status

In a Codex CLI session, type:

/status

It prints the active model, approval policy, writable folders and token usage. When you’re signed in with ChatGPT, it also shows how much of each rate-limit window is left, in lines like these:

5h limit:     [████████████████░░░░] 78% left (resets 16:40)
Weekly limit: [███████████████████░] 94% left (resets 09:12 on 14 Nov)

The exact layout changes between versions, and limits appear only once Codex has heard back from the server in that session. If the lines are missing, send a prompt and run /status again.

/usage for history

Newer Codex versions also have /usage, which shows your account’s token activity over time:

/usage daily
/usage weekly
/usage cumulative

Plain /usage opens a menu where you can view activity or redeem an earned limit reset, if your account has one. It needs a ChatGPT sign-in.

Keep limits in the footer

Codex’s footer is configurable. Type /statusline, add the rate limits item and order it with the others (model, context, git branch, tokens). The choice is saved to tui.status_line in ~/.codex/config.toml, so it sticks across sessions.

On the web

Your plan’s Codex usage and credits are also shown on ChatGPT’s Codex settings, under Usage (chatgpt.com/codex/settings/usage). Sign in with the same account Codex uses, which is the most common reason the numbers don’t match.

Read it from the logs on your Mac

Codex writes each session to a JSONL file under ~/.codex/sessions/ (or $CODEX_HOME/sessions). Its token_count events carry the same rate_limits data /status uses: primary is the 5-hour window and secondary is the weekly one. You can read the latest value without opening Codex:

f=$(find ~/.codex/sessions -name 'rollout-*.jsonl' -print0 | xargs -0 ls -t | head -1)
grep '"token_count"' "$f" | tail -1 | jq '.payload.rate_limits'

{
  "primary":   { "used_percent": 22.0, "window_minutes": 300,   "resets_at": 1762706400 },
  "secondary": { "used_percent": 6.0,  "window_minutes": 10080, "resets_at": 1763107920 }
}

used_percent is how much is gone; subtract from 100 for what’s left. window_minutes 300 is 5 hours and 10080 is 7 days. resets_at is Unix time: date -r 1762706400 turns it into a clock time. Some older versions wrote resets_in_seconds instead, counted from the event’s timestamp.

The value is as fresh as your last Codex request. If the reset time has already passed, that window has reset even though the log hasn’t been updated.

Make your limit last longer

Most of a Codex window goes on context. A few habits stretch it:

• Start a fresh session for an unrelated task instead of continuing a long one.
• Point Codex at the files that matter rather than asking it to explore the whole repo.
• Use a lower reasoning effort for routine edits and keep the highest setting for hard problems.

When you run out

Codex tells you when a limit is hit and when it resets. Your options are to wait for the window, switch to a smaller model for the rest of the window if your plan limits models separately, or add credits if your plan offers them. Running several Codex sessions in parallel draws on the same allowance, so a second terminal halves your runway.

Seeing it in the menu bar

Stat Space’s Usage limits panel reads those same session logs for Codex, at most once a minute, and shows remaining percent per window, the reset time and a pace line such as “limit in ~40m” or “on pace”. Nothing leaves your Mac for Codex. It sits in the Agents tab and as one line on the pinned Agents card, next to your Claude Code limits if you use both. It’s free and off until you enable it in Settings › Agents.

The same tab shows each running Codex session’s runtime, CPU, child processes and ports, and flags what it left behind when it ends. See the Agents feature.